Proceedings oj The SouthAfrican Sugar Technologists' Association - June 1993

THE DESIGN AND OPERATION OF BOILER PLANT UTILISINGFURFURAL RESIDUE AS A FUEL

By 10. P. NAUDE, 'P. 1. McINTYRE and 2St. 1. FIELDI John Thompson Africa (Pty) Limited, Bellville, Cape

2 CG Smith Sugar Limited, Sezela

AbstractBagasse is the fibrous waste which results from the pro-

cessing of cane to produce sugar. Furfural can be producedfrom the bagasse yielding a waste product, furfural residue,which can be disposed of in a boiler for generating steam.While bagasse and furfural residue are waste materials fromthe same feedstock, the physical properties and combustioncharacteristics of the two fuels differ markedly. Boilers de-signed for burning furfural residue incorporate features toenable stable and continuous combustion of the fuel. Op-erating experience gained over a number of seasons has re-sulted in modifications to peripheral plant to enable the safechange over from one fuel to another. Comparison with theproperties of bagasse leads to a formal approach which canbe used to describe the combustion process.

IntroductionSezela Sugar Mill is capable of crushing 440 tlh of cane

and is one of the largest mills in the South African sugarindustry. Steam requirements for the mill are supplied byfour boilers noted as follows:

NO.1: 140 tlh three-pass with dump grateNo.2: 57 tlh three-pass with travelling grateNo.3: 57 tlh three-pass with travelling grateNo.4: 130 tlh single-pass with travelling grateAt Sezela, the coarse fraction of mill bagasse is utilised in

a by-products plant for producing furfural. Steam is requiredby this plant, and as a consequence, the additional process .heat demand is made up by burning coal. Bagasse, furfuralresidue and coal were the fuels specified for generating steamwhen the No. 4 boiler was ordered. The specification re-quired that each of the fuels be burnt individually or incombination with each other.

Limited experience is available on the combustion offur-fural residue, in view ofthe limited number ofmills at whichfurfural is produced from bagasse. At the time the boilerwas designed, it was known that furfural residue had beenburnt at Victorias Milling in the Philippines and CentralRomana in the Dominican Republic.

Furfural ResiduePlant materials containing pentosans can be used as the

feedstock for producing furfural. At Sezela, the coarse frac-tion of the bagasse is fed to reactors where furfural is pro-duced in contact with steam. Paturau (1969) illustrated theproduction of furfural by the followingsimplified equationshowing the hydrolysis of Xylan to Xylose, which subse-quently loses three water molecules to form furfural:

CSHg04 + H20 ~ CSHIOOS ~ CSH40 2 + 3H20Xylan Xylose FurfuralThe cellulose molecule loses two water molecules and as

a consequence, the furfural residue has a higher carbon to

176

hydrogen ratio and a higher carbon to oxygen ratio thanbagasse. Typical analyses for both furfural residue and ba-gasse are given in Table 1, confirming the differences be-tween the proportions of carbon, hydrogen and oxygen withinthe accuracy of the analysis.

Table I

Chemical and physical characteristics of furfural residue and bagasse

Furfural residue Bagasse

Proximate analysis:Ash (%) 2,9 2,0Volatile (%) 37,2 40,2Carbon (%) 8,0 5,8Moisture (%) 51,9 52,0

Ultimate analysis (daf):Carbon (%) 56,2 48,6Hydrogen (%) 5,8 6,0Oxygen (%) 37,4 45,4Nitrogen (%) 0,5 -Sulphur (%) 0,1 -

Combustion:Gross calorific value (kJ/kg) 9800 8740Heat released per kg of airconsumed (10) 3 180 3415Theoretical maximum CO, (%) 19,9 20,6

Physical propertiesThe densities of bagasse and furfural residue differ mark-

edly. Typically, furfural residue has a bulk density of 450kg/rrr',compared with that for bagasse ofapproximately 140kg/rrr'. At Sezela, bagasse moisture contents are of the orderof 51,5%, compared with furfural moistures of around 53%.

Size gradingThe texture of furfural residue indicates a grading much

finer than that of bagasse. Typical gradings of bagasse andfurfural residue are represented in Figure 1. The bone dryresidue shown in this figure indicates two distinct size frac-tions. The first is similar to bagasse where a large proportionof the sample has a size in excess of 1 mm. The second andlarger fraction, is similar to the grading anticipated for ba-gacillo and has a size of between 105 and 185 microns. Thesize grading of wet furfural residue appears to be coarser dueto the agglomeration of the wet particles. An explanation ofthe two distinct size fractions in the bone dry furfural residuesample could be ascribed to the physical breakdown of thebagasse in the reactors, where only part of the bagasse isbroken down, with a smaller fraction remaining almost in-tact after processing.

Proceedings of The South African Sugar Technologists' Association - June 1993

31,066,2

26,980,7

Furfural Bagasseresidue

Residue 48%moistBagasse 46%moist

160 200 240 280 320 360 400 440 480Temperature (e)

50-r----------------_

40

10

Volatiles from thermal decomposition (%)Weighted mean activation energy (kl/rnol)

c:.~CIJ 30CIJ

.9CIJgj:2 2015OJCiiex:

FIGURE 2 Volatile release profiles for bagasse and furfural residueat a constant heating rate of 45C per minute in an inertnitrogen atmosphere.

Table 2

Volatile release rates

101Particle Diameter (mm)

FIGURE 1 Typical size grading of bagasse and furfural residue.

Chemical analysesThe chemical analyses noted in Table I were used in de-

fining the combustion characteristics of the two fuels. How-ever, cognisance also has to be taken of the size grading andthe matting propensity of the residue. As a consequence .stable combustion would generally only be achieved whe~the moisture content of the furfural residue does not exceed53%, while for bagasse this limit would be 55%. Bagassealsocontains more fibrous particles which increase the voidagewhich would indicate that drying of the residue would b~more difficult than bagasse.

100

~ 90c 80 ///IIIs: //~ /CIJ 70 /CIJ ' ,/OJ /

...J 60 /:E IC /'iii 50 /3::>- /al 40

cf!..//1OJ 30>~'s 20 //E / - Residue 52%moist::l - Typical Wet Bagasseo 10

--- Residue bone dry

50,-----------------------,

FIGURE 3 Combustion profiles for bagasse and furfural residue ata constant heating rate of about 50C per. minute in anatmosphere of air ..

CombustionThe sample was heated at arate of about 500 e per minute

during the combustion tests. The results of the tests graph-ically presented in Figure 3 show an initial moisture release,followed by combustion. These results are summarised inTable 3, where the temperature at onset of combustion oc-curs 200 e earlier for bagasse than for residue. The activationenergy to burn the residue is about 10% higher than forbagasse.

440400360200. 240 280 320Temperature (C)

160

-- Residue 53% moist~ Bagasse 45%moist

10

40c:.

~CIJ 30CIJ

.9CIJCIJIII:2 20's

~f1.

Analysis of Combustion PropertiesThermo-gravimetric analysis (TGA) can be used as a com-

bustion tool in analysing the combustion properties ofvari-ous fuels. The work of Raman (1981) indicated that thepyrolysis of cellulosic materials is heat transfer controlledfor particle sizes greater than 60 mm. For sizes between 2and 60 mm, they contended that it is controlled by bothheat transfer and chemical reaction, while for sizes less than2 mm, the pyrolysis is reaction controlled. Shamsuddin(1992) has described the use ofTGA in predicting the ther-mochemical conversion of biomass, and in particular, thedevolatilisation of palm oil solid wastes which are similarto bagasse. A comparative analysis of bagasse and furfuralresidue, based on TGA techniques, was undertaken to pro-vide information on differences in the combustion reactionof the two fuels. Samples of bagasse and residue were givento Falcon laboratories to undertake the TGA analyses.Volatile release rate

The rate and extent of decomposition of the fuel can be~een from the volatile release profile shown in Figure 2. Anmert atmosphere was maintained in the furnace with nitro-gen to prevent combustion. Table 2 summarises the resultsof Figure 2, where the profile can be described as follows: The initial peak represents the moisture loss. The second peak represents the volatiles loss. The initial

smaller peak in the bagasse curve may be due to the de-composition of the pentosans.

The final peak occurring at about 4400 e represents theremaining char.

The volatile curves indicate a larger volatile release at-tributed to the bagasse, with a larger char release attrib-uted to the residue.

177

Proceedings of The South African Sugar Technologists' Association - June 1993

Table 3

Thermo-gravimetric analysis from the combustion profiles

Furfural, Bagasseresidue

Temperature at onset of combustion (0C)Apparent activation energy to burn (kJ/mol)Temperature at maximum mass loss (0C)

249109369

228101366

fewer inter-connections than for bagasse. These conditionsindicate a reduced internal voidage and a greater density ofthe material as a whole.

The study shows that care is necessary to ensure efficientcombustion of furfural residue. The reduced porosity limitsthe diffusion of oxygen to the combustible matter. This istranslated into an increased air to fuel ratio to compensatefor reduced porosity when burning furfural residue than whenburning bagasse.

Design DetailsThe difference in ignition temperatures can be attributed

to the faster drying rate of the bagasse, where there is highervoidage due to the larger proportion of stringy fibrous ma-terial contained in the bagasse as opposed10 the furfuralresidue.

The differences in the intensity and ternt erature of thecombustion peaks for the bagasse sample are due to thedifference in mean particle size and volatile contents.Drying of the fuel

In the tests for measuring the volatile release rate andcombustion characteristics, the sample was placed in thefurnace at ambient temperature and then heated at a fixedrate to about 550C. During this time the sample dried,caught fire and burned. Drying took between 4 and 5 min-utes. This does not happen in normal practice.

A further series of tests was conducted with furfural res-idue with the furnace initially pre-heated to 500C and sec-ondly to 1 OOOC. In the furnace pre-heated to 500C, thesample dried out at about 430C and ignited after about 36seconds, while at 1 OOOC preheat, ignition occurred after15 seconds at 550C.

These tests show that the residue can take a significanttime to dry out when fed into a furnace. Fuel can fall on thegrate, restricting the flow of air through the fuel, causing thefuel to smoulder, and in extreme cases, leading to loss ofignition.

This supports the need for undergrate air heating to pro-mote combustion. Magasiner (1987) has reported on the cel-lulose: nature of fibrous fuels which are difficult to dry bymeans of radiant heat. In an experiment to quantify the rateat which heat is transferred to a pile of bagasse by radiation,it was noted that after sixteen minutes the temperature atabout 60 mm below the surface was less than 100C. Theseexperiments showed that drying could best take place byintroducing hot air through the grate from below, rather thanby radiation from above.

The fine nature of furfural residue and the results of theTGA analysis show that additional heat would be requiredin the combustion air to dry the furfural residue to obtaincombustion performance similar to a typical bagasse.Microscopic analysis ofbagasse andfurfural residue

A microscopic analysis ofboth the materials indicates thatbagasse is largely fibrous, open celled, thin walled and veryporous. Cell cavities are largely intact and uncompressed,with the cell walls relatively rigid. This leads to high internalvoidage and a large internal surface area, much of which iswell connected.

Furfural residue analysed under a microscope indicates amore compact structure with the cell walls less rigid. Piecesof the fibrous material are in the form of small grains inwhich the cellular cavities have been reduced. The cell wallsare unevenly broken or massed together, leading to reducednatural porosity. Internal voids are smaller individually with

178

FuelfeedersIn designing plant for burning fibrous fuels, as much at-

tention needs to be given to the handling of the fuel as tothe combustion. It was recognised that feeding of furfuralresidue to the furnace may prove difficult. A further concernwas the ability of the bagasse feeders to 'hold up' a columnof furfural residue.

Tests were conducted to prove that the traditional threedrum bagasse feeder was capable of handling the residueand ofmaintaining a head ofthis fuel. A draught was createdbelow the test feeder with a view to inducing the flow offurfural residue through the unit.

The tests showed that, with a draught below the feeder,the unit was capable of maintaining a column of furfuralresidue without by-passing. The feed characteristic was lin-ear, with a similar volumetric characteristic to that obtainedwith bagasse. The carding drum was arranged in a herring-bone pattern, rather than parallel to the shaft, to ensure thatwhen feeding very wet fuel, the bagasse or furfural residueis sliced off uniformly, rather than being fed in lumps.

Combustion chamberBoiler No.4, i.e. the 130 tlh single-pass unit, is designed

to burn bagasse, furfural residue and coal either individuallyor in varying mixed proportions. This boiler is described byMagasiner and Naude (1988) and is illustrated in Figure 4taken from their paper.

The most important component of a boiler in which dif-ferent fuels have to be burnt is the furnace. While the designrequirements for bagasse and coal are well known, data wereextrapolated for a range of fibrous fuels to describe the com-bustion characteristics of furfural residue. In particular, thesize grading and matting propensity of the residue was usedin developing guidelines for the design. Normally, for ba-gasse firing, secondary air injection is incorporated low downon the rear wall, while the fuel is introduced pneumaticallyfrom the front wall with secondary air. Additional secondaryair was introduced above the feeders on the front wall, aswell as high up on the rear wall.

The boiler was designed with a sandwich wall construc-tion. In this construction a tile is located between furnacewall tubes. The refractory tiles provide the thermal inertiarequired for stabilising combustion when burning the wetfibrous fuels, while furnace tubes provide cooling which re-duces slagging when burning coal.

OperationInitially, the No.4 single pass boiler burnt a combination

of furfural residue and bagasse. It was found that the fuelsburnt well in the furnace with a controlled fuel to air ratio.However, as operation of the by-products plant increased,so the quantity of furfural residue increased. Field et at.(1992) reported that up to about 1986 the quantity ofresidueproduced was insufficient to cause serious problems in the

Proceedings of The South African Sugar Technologists' Association - June 1993

COALBUNKER

SINGLE PASSCONVENTIONBANK

2 PASSPARALLELFLOWAIFHEATER

-=hI

--,I

FIGURE 4 Boiler designed for burning bagasse, furfural residue and coal.

i

It FIBROUS FUELt SLAT CONVEYOR

I~."~~O~~.

~1)\ I-~

PNEUMATICDISTRIBUTORS

COARSE ASHHOPPER

...

. .

mill. From that time onward 'puffs' could occur when dis-turbances caused the fuel composition, and hence the den-sity, to change rapidly. As the feederswere volumetric devices,the amount of fuel fed to the furnace was no longer in pro-portion to the air flow.

In analysing the problem, it was felt that 'puffs' couldoccur during start-up or transient operation. These two op-erating conditions can be considered separately.Start-up

If the boiler is tripped during operation with furfural res-idue, then this fuel will remain in the system and, in par-ticular, the chutes leading to the boiler will contain theresidue. Without a supplementary fuel, it is necessary toutilise furfural residue as a means of establishing stable com-bustion during the re-start. However, any start which doesnot result in the immediate ignition of the furfural residue

179

could lead to a 'puff'. Initially, it was thought that 'puffs'occurred due to wet furfural residue extinguishing the flameand then re-igniting. Current thinking is that, while the wetresidue may extinguish the flame, the puffoccurs as the resultof the dry furfural residue re-igniting as a dust cloud.

Transient operationUnstable furnace combustion conditions may take place

due to deviations from the correct air to fuel ratio. This canoccur through changes in fuel density. As the fuel feedersare volumetric flow units, changes in density may result inthe introduction ofmore fuel to the furnace than anticipated.In the particular case where the density increases due to achangeover from bagasse to furfural residue, piling can occuron the grate. Once the fuel dries, a 'puff' may occur as ad-ditional air is introduced to the furnace when the loadIncreases.

Proceedings of The South African Sugar Technologists' Association - June 1993

CombustionProperties:

Experience in the Philippines indicated that 'puffs' werenoted if the moisture content of the furfural residue fell be-low 44% (John Thompson Australia, personal communi-cation). As a consequence, sprays were used above theconveyors to damp the residue.Operation at constant density .

At the installation in the Philippines, furfural residue andbagasse enter the boiler station on separate conveyors. Aprecaution taken to ensure stable combustion was to mixthe bagasse and furfural residue in fixed proportions. Therate of production of the furfural residue was known andthe bagasse flowwas adjusted to match this rate by a variablespeed bagasse conveyor. Both fuels were mixed in a binabove the feeders and fed into the boiler by conventionalbagasse conveyors and pneumatic distributors. The propermixing of the bagasse and residue in this bin was regardedas the best means of maintaining a constant fuel density,which had the effect of reducing the formation of 'puffs'.

A further precaution was the use of a base load oil flame.The rate of oil firing was commensurate with the smalleststable flame that could be maintained. A single burner wasused with a 4:1 turndown.Fuel density compensation

It became clear that at Sezela the sudden changeover frombagasse to furfural residue could not be avoided. A 'slug' offurfural residue could be introduced to the furnace, therebychanging the density of the fuel by a factor of three, and asa consequence, affecting the air to fuel ratio to the extentthat ignition might be lost. Field et al. (1992) described thesuccessful utilisation of density measurement at Sezela as ameans of compensating for changes in density and over-coming the problem of incorrect fuel to air ratios with vary-ing proportions of bagasse and furfural residue.

The use of fuel density compensation has been extendedfrom the use ofa single density probe to a multiple ofprobeson both boilers 1 and 4. The use of more than one probeimproves the control by smoothing minor short durationchanges in density.

While fuel density measurement does enable the change-over from one fuel to another without major deviations inthe combustion conditions in the furnace, the volumetricfuel feeder is not ideally suited to the large variations indensity that occur when changing from bagasse to furfuralresidue. The result is that when feeding residue the feederspeed is slowed down and the sensitivity of the feeder isimpaired. Consideration has been given to including withinthe control logic means of reducing the number of feedersin operation when feeding high density fuel at part loads.Under these conditions, care has to be taken to ensure thatan even spread of fuel is maintained with the reduced num-ber of feeders.

Support fuelIt was noted that boilers 2,3 and 4 were less susceptible

to 'puffs' than the No. I three-pass unit. The better operationof these units was attributed to combustion support offeredby the coal bed maintained on their grates.

During a sudden mill stop, the bagasse carriers and feedchutes contain the last fuel Which, due to the high rate offurfural production, is more likely to be furfural residue.

Coal firing has been introduced on to the No. I three-passboiler, so that in the event of loss of ignition through a millstop, ignition is re-established on the boiler by initially burn-ing coal prior to introducing furfural residue. Alternatively,bagasse can be utilised as a start-up fuel.

180

Flamefailure detectionFlame flicker monitoring systems will be incorporated into

boilers 1 and 4 during the 1993 season to protect these unitsin the event of a flame-out. Precautions generally applied topulverised fuel firing have also been implemented as an ad-ditional safety measure.

ConclusionsWhile furfural residue is a derivative of bagasse, the com-

bustion properties of this fuel differ from that of bagasse.Major differences occur in the physical nature of the fuels,which affects the drying and hence combustion.

Operational experience has shown that care has to be taken,particularly during start-up and transient operation. This hasbeen overcome at Sezela by introducing a support fuel tothe boilers and by density compensation.

In assessing a fibrous fuel, it is useful to compare theproperties of that fuel with an existing fuel for which designparameters have been established. In this instance, ~urfuralresidue, when compared to bagasse, can be descnbed asfollows:Physical Properties: Furfural residue has a higher density

and a finer size grading.Chemical Properties: Furfural residue has a higher carbon

to hydrogen and carbon to oxygen ra-tio, a higher calorific value and lessheat is liberated per kilogram of oxy-gen consumed.Furfural residue is more difficult todry due to its matting propensity, andto the lower voidage of the fuel. On amicroscopic basis the fuel has a poorervoidage, making it more difficult forthe diffusion of oxygen to the com-bustible material. Furfural residue hasa higher activation energy, whichwould indicate the need for higherundergrate air temperatures.

AcknowledgementsThe authors would like to thank CG Smith Sugar and John

Thompson Africa (Pty) Limited for allowing them to publishthis paper and for providing an atmosphere to enable a betterunderstanding of the problem.

REFERENCESField, St J, McKenzie, K, Scholz, JW and Sidinile, V (1992). Fibrous fuel

density compensation in boiler combustion at Sezela. Proc S Afr SugTechnol Ass 66: 199-202.

Magasiner, N and de Kock, JW (1987). Design Criteria for fibrous fuel firedboilers. Energy World August/September 4-12.

Magasiner, Nand Naude, DP (1988). Pneumatic spreading of fibrous fuelsand coal in boiler combustion chambers. Proc S AfrSug Technol Ass62:79-84.

Paturau, JM (1969). By-Products of the Cane Sugar Industry. Elsevier Pub-lishing Company, Amsterdam, 96.

Raman, P, Walawender, WP, Fan, LT and Howell, JA (1981). 'Thermo-gravimetric analysis of biomass'. IndEngChem Process Des Del' 20: 630-636.

Shamsuddin, AH and Williams, PT (1992). Devolatilisation studies of palm-oil solid waste, by thermo-gravimetric analysis. J Ins! Energy 65: 31-34.